Flame Temperature Measurements using LabVIEW for Multi-Wavelength Emission Absorption Spectroscopy

Author(s):
Leslie E. Bauman - Mississippi State University

Industry:
University/Education

Products:
LabVIEW

The Challenge:
Update a legacy spectroscopic instrument to a modern detector and computer with a minimum of development effort and the goal of avoiding future obsolescence. The instrument should, like its predecessor, provide real-time detector control, data acquisition, analysis, logging and display of results.

The Solution:
A PC-based LabVIEW virtual instrument system controlling a CCD detector. The development effort was minimized by choosing a CCD detector provided by the supplier with a LabVIEW package of device drivers, control routines, and example routines.

Introduction
Measurement of temperature is fundamental for diagnosing high temperature flows and various spectroscopic techniques are widely used for non-intrusive diagnostic measurements in research laboratories and increasingly in industrial processes. A classical technique for measurement of flame temperatures is that of ‘line reversal’, more generally categorized as emission absorption spectroscopy. Applied to an atomic line in the visible spectrum, emission absorption provides easily implemented, accurate measurements of not only temperature (Range: 1600 - 3000K, Accuracy: ±10 K) but also the species concentration of the atomic emitter. With robust instrumentation and real time analysis, the technique is suitable for monitoring or controlling a high temperature process stream.

Legacy Instrument
The MEAS (multi-wavelength emission absorption spectroscopy) instrument was developed for temperature, atomic density and electron conductivity measurements on prototype coal-fired MHD power generators and served as a workhorse instrument for over a decade. The main instrument components are a tungsten strip lamp for the reference light source, optics for separation of emission and absorption light beams, a monochromator for spectral separation, a multi-channel detector, and computer.

Difficulties in applying emission absorption spectroscopy to industrial process flows lead naturally to the use of a multi-channel detector for the measurement. First, the line-of-sight technique provides a measurement of the average gas temperature when applied as a spectral integral over the resonance line. For the measurement to approach the hot core flow temperature in a pipe flow or plume by avoiding error caused by cool boundary layers, the measurement must be spectrally resolved on the edges of the emission line. Second, to lessen broadband absorption errors such as dirty windows, particles in the flow, optical misalignments perhaps from facility vibrations, etc. measurements must be made at two nearby wavelengths. Third, for optimized measurements over a wide range in atomic density and temperature the two wavelength positions must be rapidly adjustable. Finally, the emission and absorption signals must be taken at the same time or within a temporal spacing less than the time scale for significant fluctuations in the flow properties. All of these considerations are met by the use of an area detector scanned in two spectral strips for the separate emission and absorption signals.

The legacy instrument was integrally tied to 1980s technology: a vidicon detector (a delicate electron-beam device), an LSI-11 computer, and a graphics monitor. At the time of development, real time control, data acquisition and analysis demanded custom software. Control of the detector and real time data collection was done in low level programming (MACRO) while FORTRAN routines was used for analysis and display. The utility of the instrument and the extensive investment in software kept the instrument in use well beyond a reasonable lifetime. When the opportunity arose for measurements at Air Liquide, it became imperative to modernize the instrument.

Modern Instrument
A decision was made to modernize the MEAS instrument with a solid-state CCD detector and a PC using LabVIEW software. The instrument software was developed on a 200 MHz pentium PC, with Win98 operating system and LabVIEW 5.0. The MEAS instrument was written for any one of the Princeton Instruments (PI) family of CCD detectors. This choice was dictated by the availability of CCD detectors in the authors’ research labs coupled with the recent release of a Princeton Instruments package of LabVIEW CCD routines.

With drivers, control routines, and a few sample virtual instruments provided by Princeton Instruments (now Roper Scientific) CCD control and data acquisition was accomplished with minimal effort and no low-level programming. The PI routines were well documented and easily incorporated in building the instrument. Two important features implemented beyond those included in the sample PI routines were 1) an option for differing CCD scan orientations and 2) selection of software or hardware binning of pixel data depending upon signal intensity. These features, essential for two-track spectral scans, were easily implemented using LabVIEW array functions. Sample CCD data read/write routines were expanded to include additional header information and differing datatypes making use of LabVIEW numeric conversion and data manipulation routines. Two small bugs were found in the distributed CCD routines, perhaps not noticeable for more typical CCD data but deadly for two-track data. LabVIEW debugging capabilities, probes, breakpoints, and highlighted execution, were instrumental in finding and fixing the bugs.

The next greatest time saving in the development was in the user interface and data display routines. A run-time menu provides a user-friendly interface to the control, initialization and data collection routines as individual frames in a single LabVIEW vi as compared to several executables in the legacy instrument. Disabling (greying out) menu items helped to foolproof the operation of the instrument. Key to accurate temperature measurements, are the spectroscopic initializations: relative detector response function, reference lamp signal, background correction, etc. Routines are enabled as the initializations are successfully completed. For instance, the effective lamp temperature must be determined before the white light signals are used to determine the detector response calibration.

Graphical presentation of data is fundamental to the instrument. An intensity graph is used for viewing the full frame CCD signal in order to set the two spectral regions, XY graphs are used for spectral plots, temperature history, bar charts for histograms, etc. Figure 2 illustrates the temperature results display panel. Attribute nodes were used to tailor the various displays for a tightly configured, efficient instrument panels.
The first version of the instrument used the legacy line reversal analysis software directly as a code interface node. For increased flexibility, a complete conversion to g routines was made in the second version. With the author’s increasing competence in g programming, several previous post test data analysis routines were also incorporated into the real-time instrument such as Voigt lineshape calculations for spectrally integrated measurements. Rewriting the legacy FORTRAN routines in g was remarkably straightforward.
The one completely new feature of the instrument is internet data transfer of results for data logging on the Mississippi State University combustion test facility computer. This was implemented painlessly with LabVIEW NetDDE calls by drawing upon example client/server virtual instruments.

Industrial Measurements
The LabVIEW MEAS instrument was tested on industrial oxy-fuel (100% O2 with natural gas) burners with firing rates up to 500 kW in a pilot furnace located at Air Liquide’s research facility in Countryside, Illinois. Conducting temperature measurements on industrial scale oxy-fuel flames poses special problems compared to air-fuel combustion because of the higher temperatures and steeper gradients. Conducting these measurements with the MEAS instrument provides a means to obtain real-time temperature information on such flames. An illustration of such measurement results is shown in Fig. 4. In this case, the burner was translated vertically to map the temperature profile. Histogram temperature plots collected at various positions in the flame reveal the flame geometry and stability. Note that at -9.8 cm from the centerline the temperature distribution is broadened compared to results near the center indicating that measurement is near the edge of the flame. This information is particularly useful for validation and parameter adjustment in 3-D numerical models and flame optimization for the process application.

Conclusions
The MEAS instrument is a significantly improved version of the legacy spectroscopic instrument. Modernization using LabVIEW met the goal of minimal development effort due to the inherent capabilities of LabVIEW and the complete package of LabVIEW routines provided by the CCD detector supplier. Whether the goal of avoiding future obsolescence has been met or not can only be determined by the passage of time. However, the instrument is now packaged around industry standard software and a commonly used family of CCD detectors and will hopefully be applied for temperature measurements in high temperature processes for a number of years.

Browse All Case Studies »